## Methodology

This section outlines the data collection, processing, and cleaning methodology for the Fishmap Data project.

**1.1 Data Sources**

| Source | Dataset |
| :---- | :---- |
| Norwegian Environment Agency (TEOTIL) | Pollution data by region, pollutant, and sector |
| Norwegian Directorate of Fisheries | Fish farm sales, stock, and site statistics |
| Norwegian Food Safety Authority (Akvakulturregisteret) | Aquaculture site registry |

**1.2 Data Collection**

Input data was sourced from Norwegian government agencies and acquired via manual download. Key considerations include:

* **Regional Coverage:** Data covers all Norwegian regions, enabling regional comparison of aquaculture activity and environmental impact.
* **Time Series Consistency:** Pollution and farm statistics are tracked annually to support trend analysis.
* **Site-Level Resolution:** The salmonids registry provides site-level coordinates, species, and distance data for spatial analysis.

**1.3 Data Cleaning and Processing**

Raw data was cleaned and processed through the data-cleaning pipeline prior to ingestion into this project:

* **Pollution Data:** Extracted from the TEOTIL 2023 report Excel tables. Columns standardised and regional totals computed across pollutant types (N, P) and source sectors (aquaculture, agriculture, sewage, industry, built-up, background).
* **Fish Farm Data:** Combined sales, stock, and site datasets from the Directorate of Fisheries. Aggregated to region and year level.
* **Salmonids Registry:** Filtered from the full Akvakulturregisteret to salmonid species only. Enriched with region assignment and distance-to-coast calculations.

**1.4 Outputs**

Cleaned datasets are written to the `outputs/` directory and tracked with full data lineage via the sunstone-py library, including content hashes, execution timestamps, and field-level derivation records.